def save_snps(self):
""" Save SNPs to file.
Returns
-------
bool
true if SNPs saved to file in output directory
"""
if self._snps is not None:
try:
if lineage.create_dir(self._output_dir):
output_dir = self._output_dir
else:
return False
file = os.path.join(output_dir, self.get_var_name() + '.csv')
print('Saving ' + os.path.relpath(file))
self._snps.to_csv(
file,
na_rep='--',
header=['chromosome', 'position', 'genotype'])
except Exception as err:
print(err)
return False
else:
print('no SNPs to save...')
return False
return True
def _get_path_hapmap_h36(self):
""" Get local path to HapMap for hg18 / NCBI36, downloading if necessary.
Returns
-------
str
path to hapmap_h36.tar.gz
References
----------
..[1] "The International HapMap Consortium (2007). A second generation human haplotype
map of over 3.1 million SNPs. Nature 449: 851-861."
"""
if not lineage.create_dir(self._resources_dir):
return None
hapmap = 'hapmap_h36'
destination = os.path.join(self._resources_dir, hapmap + '.tar.gz')
if not os.path.exists(destination):
try:
# make FTP connection to NCBI
with ftplib.FTP('ftp.ncbi.nlm.nih.gov') as ftp:
ftp.login()
ftp.cwd('hapmap/recombination/2008-03_rel22_B36/rates')
# download each HapMap file and add to compressed tar
with tarfile.open(destination, 'w:gz') as out_tar:
for filename in ftp.nlst():
if '.txt' in filename:
path = os.path.join(destination, hapmap,
filename)
self._print_download_msg(path)
# open temp file, download HapMap file, close temp file
with tempfile.NamedTemporaryFile(
delete=False) as fp:
ftp.retrbinary('RETR ' + filename,
fp.write)
# add temp file to archive
out_tar.add(fp.name,
arcname=os.path.join(
hapmap, filename))
# remove temp file
os.remove(fp.name)
ftp.quit()
except Exception as err:
print(err)
return None
return destination
def _download_file(self, url, filename, compress=False, timeout=30):
""" Download a file to the resources folder.
Download data from `url`, save as `filename`, and optionally compress with gzip.
Parameters
----------
url : str
URL to download data from
filename : str
name of file to save; if compress, ensure '.gz' is appended
compress : bool
compress with gzip
timeout : int
seconds for timeout of download request
Returns
-------
str
path to downloaded file, None if error
"""
if not lineage.create_dir(self._resources_dir):
return None
if compress and filename[-3:] != '.gz':
filename += '.gz'
destination = os.path.join(self._resources_dir, filename)
if not os.path.exists(destination):
try:
self._print_download_msg(destination)
if compress:
open_func = gzip.open
else:
open_func = open
# get file if it hasn't already been downloaded
# http://stackoverflow.com/a/7244263
with urllib.request.urlopen(url, timeout=timeout) as response, \
open_func(destination, 'wb') as f:
data = response.read() # a `bytes` object
f.write(data)
except Exception as err:
print(err)
return None
return destination
def save_snps(self):
""" Save SNPs to file. """
if self._snps is not None:
try:
if lineage.create_dir(self._output_dir):
output_dir = self._output_dir
else:
output_dir = ''
file = os.path.join(output_dir, self.get_var_name() + '.csv')
print('Saving ' + os.path.relpath(file))
self._snps.to_csv(
file,
na_rep='--',
header=['chromosome', 'position', 'genotype'])
except Exception as err:
print(err)
else:
print('no SNPs to save...')